Mesh refinement strategies for capacitance extraction based on residual errors

We investigate mesh generation for capacitance calculations. The refinement scheme is controlled by an error estimator based on the residual of the solution calculated on a coarse grid. Numerical experiments are presented. The surface gradient of the residual provides a mean to estimate the discretization error. The grids produced by our algorithm allow accurate estimates of the capacitance with significantly fewer panels than uniform discretizations. Adaptive procedures that minimize the discretization in other norms, like the energy norm, could result in even better grids